EP0789310B1 - Intelligent CAD method embedding product performance knowledge - Google Patents

Description

This invention relates to computer aided design, and more particularly to computer systems that embed product performance knowledge as a rule in computer aided design software programming.

It is conventional to use a find and fix mode to correct product designs after they have been placed in use. This is exemplified by computer aided drafting where the product design is created by instructions given to a computer via a designer's knowledge as input data; the computer aided drafting process organises the input data and provides a deliverable in the form of an executed drawing. Field experience will tell if there is a problem due to factors such as environment, durability, customer satisfaction, damageability, recyclability, manufacturability, cost, features, finish, fit, and other internal and external influences. It is desirable to eliminate such problems up front during the design development process; the process should make the deliverable insensitive as much as possible to the aforementioned factors subsequently encountered.

Conceptual approaches that attempt to store knowledge of a product's performance fail to embed such knowledge in the rule making process that is an inherent part of the design development process. In such a storage system, a separate software product performs as an interface to the design process and asks questions of the designer who is the storehouse of information; the process then draws a design from known components based upon such additional information to create only an initial design from known components.
None of these concepts integrate updated product performance knowledge or specifications into the design development process to morphically and recursively change the design. The unknown design should evolve by continuous iteration based on cross-assessments of design influencing factors.

EP-A-0561564 discloses a knowledge-based artificial intelligence system which provides design advice. The artificial intelligence system includes a knowledge base of design information. Users of the system indicate an area about which they require design advice. The system provides the relevant advice. Included in the advice is an indication of the 'owner' of the advice. The advice and the relationship between the design made by the user are part of a trace of the users' session with the system. The trace becomes part of a design document for the design. When the design is reviewed, the trace is reviewed as well. The system includes an interface for updating the knowledge base, and if the design review indicates a need to correct the knowledge base, the corrections are made using the interface for updating.

US-A-5297054 discloses an automated generative gear design process, which designs parallel axis gear sets to meet constraints and performance goals. The process uses state space searches, generate and test, and other knowledge based techniques to automatically generate gear set designs and recommend cutting and inspection tools. Specific knowledge applied includes standard gear and gear set equations and conditions for use, standard and specially developed gear set design methodologies, models of generic gears and gear sets, the specific limitations of geometry of various gears and gear sets, the ability to evaluate designs based on performance relative to goals, characteristics of available cutting and inspection tools, lubricants, and materials.

According to the present invention there is provided a method as set out in claim 1.

A method embodying the invention embeds product performance knowledge as a rule or as geometric parameters into a design and development process software program that not only stores knowledge from the development process but provides for two-way associativity influencing the design. Rules, parameters, associativity, relations, and knowledge generated by the process are used to revise and improve an unformed design as it is morphically developed. An intelligent CAD process is provided that is capable of morphically changing the design shape of a part based upon its relation to other parts and other external influences.

The invention will now be described, by way of example, with reference to the accompanying drawings, in which:

Figure 1 is a schematic diagram of a conventional computer aided design process;

Figures 2 and 3 are schematic representations of how the Intelligent CAD system of this invention;

Figure 4 is a schematic representation of a cantilevered beam to be designed illustrating embedment of associativity principles;

Figure 5 is a flow diagram illustrating use of the intelligent CAD process of this invention to evolve a design of a motor vehicle bumper using embedded subprocesses to revise and improve an inchoate design of the bumper; and

Figures 5A-5D depict subprocesses of figure 5.

As shown in Figure 1, unorganised human data 10, unorganised machine data 11 and previously stored data 12 are put into a computer gathering process 13; the gathering process calculates, makes calculations, observations and measures to provide organised knowledge 14. In the case of computer aided drafting, the human data 10 would be a designer's instructions based upon skill and prior work steps performed by the designer. The stored data 12 would be previous prints, drawings, layouts and geometric data generated by algorithms. The stored data may also include features: rounds, fillets, bosses, slots, customising aspects, or parametric (indefinite) dimensions. The organised knowledge 14 can be a three-dimensional data file having a viewable geometry displayable on a raster or vector device.

The invention, as shown in Figure 2, not only goes through the gathering step 13 as described in Figure 1, but additionally goes through an organising step 9 that provides script as a deliverable along with data 16 that together define an inchoate design. The script is an executable set of instructions created by the organising step. Such script and/or data is then subjected to an iterative editing process 17 (manual or machine) using special stored information 18a or 18b to affect the inchoate design and eventually produce an executable 19 in the form of a viewable geometry of the part as well as new script. The editing process 17 may take place in several subprocesses 20, 21 ,22, etc., each dedicated to a specific function that responds to new current information on such function such as illustrated in figure 3.

The example of a part to be designed, as shown in Figure 4, utilises a cantilevered beam 23 supported on a fixed surface 24 prescribed to support a load 25 with beam deflection not to exceed X. The beam length is to be no less than L; the diameter is to be no greater than D, and the material is aluminium (an environmental consideration). The beam is not to contact part B and the clearance Y is not to be encroached. Loads may be associated (related) to an arbitrary set of geometric entities possessing parameters and features; in the case of the beam this will be a set of points, lines, surfaces, etc. and the sets will contain subsets. There are several associative considerations: part to part, part to environment (non-geometric), geometry to part, etc. When loads are assigned to geometric entities, the geometry becomes associated to the environment. Relations may be written whereby environmental factors influence the design criteria, e.g. dimensions. Thus, using subprocesses for consideration such as manufacturability, costs, all constrained by such factors as size, section properties, area, and volume, the subprocesses will provide a two-way influence on the design by precipitating instructions which in turn executes the process which thereby influences the design.

Turning now to engine 5, the flow process for carrying out intelligent CAD design of a vehicular front bumper is laid out in broad terms.

The broad steps involve (30) creating an initial front bumper design, (31) optimising the initial front bumper design (32) generating the protection zone layout, (33) building and testing prototypes, and (34) correlating finite element analysis results to the physical tests. A breakdown of subprocesses under these broad steps to carry out the intelligent CAD development design is given below.

1. Create an initial Front Bumper Design

.1 Determine the Initial vehicle package constraints

.2 Determine the initial bumper system design objectives

.3 Generate alternative Bumper system designs

.1 Generate initial Beam and energy absorber designs

.1 Calculate maximum allowable packaging volume in car position

.1 Determine the height of the beam in car position

.2Calculate the width of the bumper system

.3 Calculate the depth of the bumper system at the centreline and frame sections

.2Select a Beam section

.1 Determine the constraints on the beam geometry due to packaging and performance testing

.2 Select a Beam section from the Beam section library

.3 Check if the beam section supports the required dimensions

.4 Calculate the range of acceptable values for each topological parameter

.3 Test the smallest beam section

.1 Calculate the required geometry and dimensions of the energy absorber

.2 Select and energy absorber from the Energy Absorber library

.3 Check if the energy absorber supports the required dimensions

.4 Select the type of rail support

.5 Check the Structural integrity of every combination of energy absorber and rail support for that particular section and its assigned dimensions by means of a lumped mass analysis

.4 Test the largest Beam section if necessary

.5Optimise the beam section if necessary

.6 Create a list of optimised bumper system designs

.2Determine their associated cost and weight

.3 Test if cost and weight objectives were met

.4 Determine if and how styling and/or vehicle constraints can be relaxed

.4 Choose the best bumper system design

2. Optimise the initial Front Bumper Design

3. Generate the Protection zone Layout

.1 Calculate the system stroke for each pendulum impact location

.2 Calculate the stroke zone

.3 Generate the Protection zone layout

4. Build and Test Prototypes

5. Correlate Finite Element Analysis Results to Physical Tests

Subprocesses 1.1 and 1.2 respectively comprise determining the initial vehicle package constraints (35) and determining the initial bumper system designs (36). This will require considerable input of human generated information and techniques, as well as machine generated information and rules. For the bumper design development this will constitute cost and weight targets, styling requirements such as whether the bumper will have a certain type of curved shape and any styling theme facia for packaging constraints, radiator information that may require cooling slots in the bumper for the air conditioning condenser as a constraint, same information that may require the bumper to be capable of sustaining multiple 5 mph collisions as well as 30 mph crash or not violate a mandated approach angle, and other engine or package constraints, such as not violating an overall vehicle length.

The inputted or gathered data is then processed to generate a bumper system design 37. First, the data is processed to produce processed knowledge containing script without a part design. This may involve calculating the maximum allowable packaging volume in the car position (38), see figure 5A, using input data as to performance requirements and vehicle rail span. This calculation subprocess may be broken down, as shown in figure 5B, into determining the height of the beam in car position (39) and calculating the width of the bumper system (40); the height width decision is fed along with styling and determining a tentative design 42 for the bumper system at the centreline and frame sections which renders a maximum allowable depth of bumper system in the car position.

With this tentatively processed information, a selection of a beam section is made (43) by the operator (see figure 5A) which begins iterative editing of the processed knowledge to continuously improve the tentative design or script. Step 43 is expanded upon in figure 5C and involves recursively editing by associative techniques, relations and features, such as the step 44 for determining the constraints on the beam geometry due to packaging and pendulum impact dimensions, using bumper system packaging information and impact barrier test requirements for beam height, depth and sweep. At the same time the processed information is edited by using input data that is knowledge-based rules and instructions, such as from a library 46, and which rules and instructions may incorporate recently learned data. With input data from the sections library, a selection and range 45 is checked to support the required dimensions (48). Then the range of acceptable values for each topological parameter is calculated (49) to produce a unique section 50 with associated manufacturing process and material with a defined range of acceptable topological parameters.

The unique section 50 is then tested by rules and instructions for every possible energy absorber and rail support, first as to the smallest section (51) and then as to the largest section (52). The testing of the selected beam section is expanded upon in figure 5D. A selection of the energy absorber is made from library (53) and then calculation of the required geometry and dimensions of the energy absorber is made (54). The results are checked to see if energy absorbed supports required dimensions. If the results fail, a new energy absorber is selected and this subprocess recursively carried out. If the results pass, a rail support is selected (55) and, along with beam section information, performance information, and acceptable range of beam sweep values, is checked as to structural integrity of every combination of energy absorber and rail support (56). This is lumped mass analysis to produce small section result 57. The same procedure in figure 5C is followed for the largest section (52) if necessary and the passed result (58), along with result 57, is fed to step 59 (figure 5A) for optimising, if necessary, the section design based on input of optimisation strategy rules and information 60. A list 61 of optimised bumper system designs is accumulated.

The feasible bumper system designs 61, including energy absorber is then analysed as to meeting associated cost and weight (62) and then tested as to whether the objectives have been met (63). If not, a determination is made as to whether styling and/or vehicle constraints can be relaxed (64). If yes, the modified information is sent back for recursive use in the process; if not, the design is deemed not meeting the targets. This completes step 30 of figure 5. The other basic steps 31-34 follow in the order listed.

As explained above, the human interactive design process of the Intelligent CAD system triggers machine driven design subprocesses. The method may spawn one or more machine driven (non-human) processes involving calculation of the design worthiness, such as an initiation of an analysis (finite element method of structural analysis) for simulation of a 5 mph pendulum impact test; concurrently, a spawned machine driven process may generate another finite element analysis for the purpose of assessing 30 mph crash worthiness. Additionally, a design optimisation executable may be spawned along with a mould flow analysis and a fatigue analysis while, at the same time, another executable is calculating material, piece and assembly cost. All these subprocesses (and more) may be initiated sequentially or concurrently and generate design knowledge never before available. Such knowledge is fed back, not only to refine the evolving design, but to add to the data base of accumulated knowledge which, in effect, becomes an ever increasing repository of bookshelf knowledge. This knowledge may be retrieved for future designs.

Claims (5)

1. A method in a computer of evolving and dynamically adapting a computer aided part design having unknown initial form, comprising: (a) gathering data related to said part consisting of (i) human generated information and techniques, and (ii) machine generated information or rules; (b) processing the gathered data by instructions to produce processed knowledge containing script defining an executable set of instructions without a part design; (c) iteratively and recursively editing the processed knowledge to continuously adapt and improve the script to form an initial executable part design by inputting into the editing process (i) recently learned data, (ii) associative techniques, relations and features, and (iii) knowledge-based rules or instructions whereby product performance information is included; and (d) inputting the editing history of step (c) into processed knowledge, such inputting creating captured improvement criteria that is iteratively and recursively fed back into either the recursive editing process or into the processing of step (b) to effect evolvement of a different executable design form.
2. A method as claimed in claim 1, in which step (c) is carried out by inputting market acceptance information, product performance information, and updated targets for weight and costs.
3. A method as claimed in claim 1 or 2, in which said associated techniques comprise relating the part design to adjacent parts of an assembly.
4. A method as claimed in any one of claims 1 to 3, in which said knowledge-based rules and instructions are associated with a library of design elements and sections.
5. A method as claimed in any one of the preceding claims, in which said iterative and recursive feeding back of said captured improvement criteria is carried out by a computer.

Images